Korean Subjecthood Revisited: Honorific Agreement and Plural Copying

The study re-examines agreement-based subject diagnostics in Korean by investigating Honorific Agreement (HA) and Plural Copying (PC) across Single Subject Constructions (SSCs) and Multiple Subject Constructions (MSCs). Using generalized linear regression and random forest modeling to analyze magnitude estimation data, the study compares the relative effects of agreement alignment and sentence type (i.e., MSC vs. SSC). The regression results showed that HA exhibited a significant alignment effect in SSC environments, while this effect weakened in MSC contexts, suggesting a sensitivity to structural configuration. PC, however, displayed limited alignment sensitivity even in SSCs and showed only marginal effects overall. The random forest analysis further revealed that sentence type is the strongest predictor of acceptability for both diagnostics, whereas agreement alignment contributes comparatively less. These findings imply that agreement-based diagnostics do not uniformly track grammatical subjecthood across constructions. Instead, their distribution is systematically conditioned by sentence type, indicating that subject-related properties in Korean are construction-sensitive.